Artificial neural network modeling of tablet coating in a pan coater

Abstract

Our study decided to use the new and revolutionary approach in the field of pharmaceutical coating processes called the artificial neural network (ANN) by using the neural networks toolbox derived from the Matlab® software. The experiments were performed using tablets of Alfuzosin Chlorhydrate as a model filler, and an aqueous solution of Surelease as a polymer in different contents. The various parameters that can affect coating thickness, weight gain, and the coefficient of variation CV, such as spray rate, air pressure, solid content, speed of the drum, pan loading, and time of coating, were studied. The properties of the coated tablets were evaluated using the ANN, and both the parameters of the coating process and the properties of the coated tablets were used as a basis for optimization, as well as the choice of the optimal structure of the ANN model. It was found that the best neural network architecture had 7 neurons in the hidden layer, with a mean square error of 3.515 and a determination coefficient of nearly 1. The relative importance of each independent variable was quantified using the Garson equation. In this study, spray rate was found to have the highest impact on the properties of tablets.